|
MicrostockGroup Sponsors
This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.
Messages - Uncle Pete
Pages: 1 2 3 4 5 6 [7] 8 9 10 11 12 ... 180
151
« on: February 10, 2024, 11:37 »
So now we need to shoot a lot of editorial.
and get paid ten cents for doing it 
And with all the hype, I've seen no numbers increase in Editorial and News income or downloads. You're right, I see loads of dimes. The fusion of Splash, a respected legacy brand, and the Shutterstock Newsroom, an elevated premium service offering, will allow us to scale our entertainment vertical. With this union, we are positioning Shutterstock as the market leader for entertainment, providing our global customer base with best-in-class editorial content. Now: "As one of the world's largest entertainment news agencies for newsrooms and media companies, our acquisition of Backgrid further elevates Shutterstock's position as the premier provider of best-in-class entertainment and editorial content," Maybe INDYCAR and NASCAR aren't entertainment?
152
« on: February 10, 2024, 11:22 »
Between 200-300 a day to have a decent amount of earnings by the end of the month
Seems excessive. I struggle to see how you can effectively create quality content at that scale unless you have a large team. More is not more.
-Mat
Yes, Less is More. I've always been unconvinced that just uploading high numbers of similar or sets, would make any difference in "decent earnings". And if the more is, more variety or different, high quality images, then more is "more". But I'm just saying, numbers alone, or portfolio size, isn't what creates income or more sales. 8
Or ∞ Somewhere between those numbers?
153
« on: February 09, 2024, 11:23 »
The thieves make money because they skim off better content. We all know that, as a general rule, 20% of the images make 80% of the money, but if they are selecting the best images, then they can certainly increase the sales percentage in their favour. It's always been quality over quantity, so if you steal 50 spectacular images, then they are likely to make more than a beginner with 50 snapshots of his back yard.
Valid points. I remember on the old SS forum, a thief was discovered and quite a few of us contributors were browsing through his port of stolen goods. The images selected were extremely high quality. Like the best of the best with all different kinds of subjects. It was quite a decent sized port too if I recall correctly. I wouldn't be surprised if this thief was making a ton of money.
I looked at the first five, 404 error. Either they are gone already or something is different on the Spanish site from the English site? But none of the checked links is anything.
154
« on: February 09, 2024, 11:15 »
I received the answer today, thank you Mat.
In my example, the MGM logo is from the fifties. Today's logo looks different, so it was wrongly assumed that this image was generated with AI. The image is now online.
I think this can also be applied to rkz91's example.
The question is how to deal with this realization in the future. Some of my images are affected.
There you go, something I never thought of. AI thinks that altered logos are AI. Good Luck.
155
« on: February 08, 2024, 12:11 »
Digitally created or manipulated versions of trademarked logos or other brand content
That's my guess for each of them.
This picture was taken on glass for reflection. And chip was held by black plastic behind it. There was fractional removal of that plastic from picture for aesthetics, but it does not change the point of picture. It is not manipulation. By the way there were others photos, where was nothing removed from it. Just simple contrast/color edit. But still rejected for same reason... 
Just a guess. The reviewer might see this as a Highly Manipulated Image. In the sense that the Logo is artistically reflected with a chip added. Is this an actual Broadcom chip? Is that the actual Broadcom, unaltered logo?
I'm not being critical, it's nice work. I'm just pointing out, that an artful image, using a name and logo with a artist composition, isn't Illustrative of the actual product or company brand name. My impression of illustrative editorial is it's supposed to be the actual items, not creative interpretations.
I get your point. This chip is from Intel, but still it was rejected with Intel logo in background as well. But Should I somehow to prove for reviewer that it is actually brand made product? lol it is non sense in my opinion Most microchips look similar
Mine was more of a personal opinion and conceptual reason why I'm guessing at the rejection reason. I have no knowledge beyond what the Adobe site says. I think that anyone, not just you, that does creative versions, and such as close cropped or altered icons or logos, might get a rejection for that reason. I'm thinking that only original logos, trademarks, icons, Etc. (seems with the exception of social platforms) is allowed. Artistic interpretations are not allowed. But until Mat says what Adobe says, I'm just writing what I think the cause was. And I have stayed away from editing for art, color, design, or improvements, when uploading anything that's Illustrative Editorial.
156
« on: February 08, 2024, 11:56 »
No I don't pay for the NY Times, sorry. But yes, all of the points are going to be decided in the future and should be interesting. Oh good, I found a free version: https://artdaily.cc/news/161538/As-fight-over-AI-artwork-unfolds--judge-rejects-copyright-claimthaler's rejected case was provocative =- claiming the ai software was the creator.
Someone had to start the ball rolling. With no case law, there wouldn't be any review or appeals and it would probably just go down as, USCO says so. Because that's the way they interpret the laws. There needs to be a challenge to create change.
157
« on: February 07, 2024, 14:21 »
Digitally created or manipulated versions of trademarked logos or other brand content
That's my guess for each of them.
This picture was taken on glass for reflection. And chip was held by black plastic behind it. There was fractional removal of that plastic from picture for aesthetics, but it does not change the point of picture. It is not manipulation. By the way there were others photos, where was nothing removed from it. Just simple contrast/color edit. But still rejected for same reason...
Just a guess. The reviewer might see this as a Highly Manipulated Image. In the sense that the Logo is artistically reflected with a chip added. Is this an actual Broadcom chip? Is that the actual Broadcom, unaltered logo? I'm not being critical, it's nice work. I'm just pointing out, that an artful image, using a name and logo with a artist composition, isn't Illustrative of the actual product or company brand name. My impression of illustrative editorial is it's supposed to be the actual items, not creative interpretations.
158
« on: February 07, 2024, 14:00 »
What is wrong with this picture, for example, by definition, no people, brand bold in the foreground.
Images that feature tight crops of copyrighted or trademarked material,
That's my guess for each of them.
Maybe, but last year this stuff was accepted. I hope that Mat can shed some light on the matter.
He's the one!
159
« on: February 07, 2024, 13:57 »
I don't see how any agency has a right to charge us commission on our own money, just for processing the claim.
Exactly! The payback isn't from additional sales but something meant to go directly to the artist. For iStock to take most of it seems wrong - it's like the owner of a restaurant stealing their employee's tips.
I also got a payback directly from DACS which was more than the iStock gross amount - it is all very confusing.
It's more like that than just the concept. It's also taking an earnings transaction and skimming 85% off the top, just for transferring it to us? There's something very wrong if IS is in fact taking 85% of some outside earnings, not the sales commissions, that they pay us 15% and do the licensing.
160
« on: February 07, 2024, 13:50 »
What is wrong with this picture, for example, by definition, no people, brand bold in the foreground.
Images that feature tight crops of copyrighted or trademarked material,
Digitally created or manipulated versions of trademarked logos or other brand contentThat's my guess for each of them.
161
« on: February 07, 2024, 13:27 »
There are a lot of people wo are selling 120 images of dogs for $1.50 and seems to be making a lot of sales. Those images are not processed well, but just rendering them, converting them , zipping and uploading + customer emails - scaring me off from trying it myself.
Anyone tried it successfully?
I haven't but I have seen just what you say and they list sales, and they appear to be successful. I see sets that are nothing but a large selection of right click images, zipped and then sold for a low price. No I haven't done that and I won't. I tried to make some high quality sets of specialized images, high resolution. No market. That's the way it goes. Crafters and scrapbook people just want bits and pieces, not high res, high quality.
162
« on: February 07, 2024, 13:21 »
yes, that's my biased opinion - your 'biased' is my ' intelligently informed opinion'
Thanks for understanding that I'm also intelligently informed and just came to a different biased conclusion.
Two very simple points, as the laws are now.
1) Only natural persons can be considered inventors
2) AI creations from scraping the web for training, are fair use.
For #1 copyright will be rejected, no matter how intricate the whole creation process may be, or whether it can be detected as AI or not. #2 the current laws defend the right of the AI companies to look at images and use that to create training.
For myself, on the other side, I still say, we added the captions, or some human did. (unless someone uses AI keywording) The text, description and keywords, included in the images, is our work and individual writing, which can't be used for training as fair use. That's my data and I own the right to that. The keywords are being illegally appropriated.
detail #1 - inventions can't be copyrighted but can be patented, so the question comes back to creation. i spent an afternoon earlier and my initial prompt didnt give what i asked for, but eventually used it to springboard & eventually ended up with 6 distinct series - it was not a simple prompt = final image; instead using a tool to refine whhat the tool gave me
individual keywords can be copyrighted,. and since the training only uses individual words, there's no violation.
the main claim for fair use is that by scraping billions of images, no individual image is copied. instead each image is broken down to tokens and there's no way to reverse the process to identify where a token came from
The word inventor applies to art, music and everything, in this context, not just "inventions". Think of it as creators or creations. I didn't write that, I quoted it. Natural persons excludes machines, monkeys, and everything else, except humans. Yes, we can't copyright a single word or idea. But a set of words, like keywords are not the same as one word. You can't copyright a recipe, but you can copyright a book of recipes, as a collection. A set of words is not just a word, or no book could be protected, because it's just a bunch of words. My argument would be, that the description and keywords are ours and applied to a specific image. By scraping the images, they are just viewing them, but by using the associated words, that we created, to describe the image, there's something beyond fair use. (and yes I admit it's a stretch) "each image is broken down to tokens and there's no way to reverse the process to identify where a token came from" Yes, and that's why we can't get credit for each new image. There is no one to one relationship from the original images to connect to the AI creations. Once trained the AI is on it's own. I think Thaler intentionally went this way and more than once, has challenged, to cause a decision and bring Machine Learning into the court system as a test case. No one would try to register an art work as created by a machine, when they know the laws say, that can't be approved.
163
« on: February 06, 2024, 12:43 »
They've been doiing it for a few years: there's a clause in our contract which can be interpreted to let them do it without an opt-out, though like so much in the contract - which is very wide in their favour - very narrow in ours, I for one didn't foresee this.
Only Exclusives? I get my DACS claimed through Alamy. They would be double dipping? I don't know, but it seems odd and confusing.
I'd certainly rather get 40% than 15% of my DACS
I don't see DACS on iStock, I guess that's the point. The other is obvious and I'd agree. 40% is much better than 15%. Since the new system at Alamy and I didn't make the $250 minimum for Gold level, I wonder if they will give me 20% of my DACS payment and grab the rest. But the question is, how can IS claim DACS if I'm already getting it from Alamy? I'd claim my own, but last time I did, the bank transfer fee was nearly more than I got, so useless to bother. I don't see how any agency has a right to charge us commission on our own money, just for processing the claim.
164
« on: February 05, 2024, 15:40 »
They've been doiing it for a few years: there's a clause in our contract which can be interpreted to let them do it without an opt-out, though like so much in the contract - which is very wide in their favour - very narrow in ours, I for one didn't foresee this.
Only Exclusives? I get my DACS claimed through Alamy. They would be double dipping? I don't know, but it seems odd and confusing.
165
« on: February 05, 2024, 15:38 »
This is modern slavery without any transparency. I
As unfair as our relationship with the agencies may be, it should not be compared to slavery. If we do not like the arrangements, we can just walk away and take our images with us. Slaves do not have this option.
But slaves could run away and that's what DP artists should be doing after this news.
"Depositphotos distributes the Files to third parties under various types of paid and unpaid agreements. Such agreements may cover various areas of permitted use, including right to use the Files for the development and training of algorithms, networks, technologies and solutions."
does not sound like some concern over lawsuits to me, but pretty much, as it says, that they will give our images to third parties that may use them for AI training. Even without compensating us.
That and the laughable, artists must agree that they cannot file a class action suit against DP. I wonder if that's even legal to include. But that part is for sure, protecting DP from lawsuits.
166
« on: February 05, 2024, 15:29 »
I'm surprised that anyone would buy a download of prompts. It just seems they are already done, and anyone can copy or use them. What's the value in making something that's already made and can be duplicated by anyone, with the same prompt? AKA, Sliced Tomato isolated on white.
rather than the text part, looking at the commands (stylize, aspect, etc) help create the look you want - but these are easily available on searches.
So are the public domain image sets that some people sell on Etsy. I guess it's buyer beware, if the buyer is dumb enough to pay for something that's already free, someplace else? We could only do worse, buying something at Amazon, for double the price at eBay or off the shelf?
167
« on: February 05, 2024, 15:23 »
yes, that's my biased opinion - your 'biased' is my ' intelligently informed opinion'
Thanks for understanding that I'm also intelligently informed and just came to a different biased conclusion. Two very simple points, as the laws are now. 1) Only natural persons can be considered inventors 2) AI creations from scraping the web for training, are fair use. For #1 copyright will be rejected, no matter how intricate the whole creation process may be, or whether it can be detected as AI or not. #2 the current laws defend the right of the AI companies to look at images and use that to create training. For myself, on the other side, I still say, we added the captions, or some human did. (unless someone uses AI keywording) The text, description and keywords, included in the images, is our work and individual writing, which can't be used for training as fair use. That's my data and I own the right to that. The keywords are being illegally appropriated.
168
« on: February 04, 2024, 12:20 »
Seems the Adobestock 1099 values are based on $$$ of record at the end of each calendar month (summed to one yearly value) and not on how/when the contributor takes their payouts?
That's the way I've understood it to be, for many years. I make a payout request around Dec. 26th each year, so the numbers are close, depending on which accounting method is used. I don't see earnings that I don't have as earnings. What if someone doesn't make payout, for four years, like DT. Are they supposed to pay taxes on the earnings, that they might never receive? I use the numbers of what any agency actually paid me, during the tax year. But yes, it appears that Adobe counts earnings, not payments to the artist.
169
« on: February 04, 2024, 12:12 »
I'm surprised that anyone would buy a download of prompts. It just seems they are already done, and anyone can copy or use them. What's the value in making something that's already made and can be duplicated by anyone, with the same prompt? AKA, Sliced Tomato isolated on white.
170
« on: February 04, 2024, 12:05 »
in an often cited case, they rejected a copyright when the owner oddly claimed the ai was the creator and he owned the copyright as a 'work for hire' - the office rejected these arguments, which is not the same as saying an artist cannot copyright a work they CREATED using a computer program.
https://www.copyright.gov/ai/docs/us-cross-motion-for-summary-judgment.pdf
the core of the decision was the Office will not register works produced by a machine or mere mechanical process that
operates randomly or automatically without any creative input or intervention from a human
author. and the applicant undermined his case by declaring the work was created autonomously by machine.. this judgment was different from declaring ai assisted works cannot be copywrote.
my emphasis shows the issue that remains to be directly litigated or legislated.
Yes, but you are biased towards one side of the decision and cherry-picking your defense. https://www.goodwinlaw.com/en/insights/publications/2023/08/insights-technology-aiml-thaler-v-copyright-register-gen-aiSo I'll do the same: When an AI receives solely a prompt from a human and produces complex written, visual, or musical works in response, users do not exercise ultimate creative control and the resulting work is not copyrightableWhen a human selects, arranges, or modifies AI generated material in a sufficiently creative way, the work may be copyrightableYes, AI work can be copyrighted, if a human selects, arranges or modifies... Thaler could have said, he edited the output and then what? Courts in the United States, Europe, Australia, and New Zealand, have all held that only natural persons can be considered inventors. Yes there is investigation and there will be more cases, but I don't think this is going to be as simple as, someone typed the prompt, so it's something that can be protected. Will AI chat text and articles be protected, because someone asked for the subject? My best argument in favor of our work being protected would be, I generate the prompt, and there's an image, which, after that, I the human, edited and altered into a new work, which therefore means I can protect my human creation, based upon an AI, public domain image.  A guiding human hand is necessary for authorship, and I am that human.
171
« on: February 03, 2024, 12:24 »
Did anyone still have sales in January'24?
I didn't have a single sale for the first time (usually around 15 a month) and I'm wondering if they're being shut down by Shutter for good.
Yes, I still had sales in January and also in February.
But it was just 7 sales in January. I had 37 sales in January 2023, just for comparison.
With the inability to submit new content I don't really see this agency going forward. Obviously with the lack of new content customers and therefore sales are also declining, as was to be expected. But even before all that my income from Bigstock was so small, that it was hardly worth mentioning. I think apart from Vector Stock they have always been the agency with the lowest income for me, so it's not one I will cry about when they finally shut down for good.
Good point. I forgot! "While we explore future improvements, starting on June 30, 2023 Bigstock will no longer accept new content submissions." Followed by, we invite you to join Shutterstock. https://support.bigstockphoto.com/s/article/uploading-submitting?language=en_US14 people here say it's a .5
172
« on: February 03, 2024, 12:12 »
If you want to erase objects in the generation process, you can do two things:
Use the --no parameter:
In your case try --no orange color
Use negative weights:
In your case try ::orange -0.5 or -0.25
Good advice, plus, in "A photograph of a diverse business team in a modern office, multi-ethnic group brainstorming around a table with laptops and digital tablets. Bright, airy office with large windows and cityscape view. Vivid colors, but not orange, Created Using: digital photography techniques, wide-angle lens, candid expressions, contemporary office design, urban background, high-resolution, professional attire" the AI sees the word Orange, and doesn't understand, but not orange. It does understand something like (orange:.15) if that's a valid command for what you are using. and should see No Orange.
or positively identify the color - in recent session, i changed prompts from 'red and gold' to purple and silver' and all images were correctly colored'
Yes, another good tip, if there's no color, the AI decides. If someone specifies a color, that will be stronger. I don't see how this is diverse, as there are nothing but people of color? No Latino, no whites, no European South, no Asians. So much for AI knowing what diversity really means. 
173
« on: February 02, 2024, 12:23 »
I'm just going to ignore the trash talk that some people throw at others, and move on.
174
« on: February 02, 2024, 12:16 »
Mat,
Is the Adobe "vendor number" the same as our "Member ID" on our contributor account page?
Thanks
Mine was.
175
« on: February 01, 2024, 15:14 »
There is basically a prompt syntax for stable diffusion.
- Type of shot (portrait, full body, from behind, macro, etc.)
- Description of the subject
- Possibly control of the depth of field by means of aperture specification or e.g. "bokeh", "shallow depth of field". Mostly with f1.8 or f2.0. Focal length of objects can also be specified, but in my opinion it has no great influence, rather directly control the portrait type such as Extreme Close Up, Close Up, Medium Shot, Full Shot, etc.
- Additional information e.g. on foreground, background, colors, (rarely also composition, image harmony)
- For portraits, additional information on the facial expressions / gestures of the person (smiling, laughing, sad, angry, etc.)
Weighting works via bracketing.
Square brackets [] reduce the weighting, round brackets () increase the weighting.
Commas and periods play a role in separating information and content. Commas are light and periods are strong separators. However, this is actually very similar to normal punctuation when writing texts.
The same applies to the negative prompts.
With LORAS (fine adjustment models), additional weighting can be applied, e.g. with<lora:sdxl_photorealistic_slider_v1-0:1> the number after :
In addition, the images can be further fine-tuned via the number of steps (iterations) and the guidance scale.
With higher settings, the images generally become more detailed, but the prompts must be formulated very precisely.
Basically, I don't think much of controlling the type of image in the standard model by specifying "photorealistic", "artistic", "surreal", etc.
Instead, you should use this in combination with specially optimized models (see https://civitai.com/models).
I believe that open source Stable Diffusion can beat Midjourney in the long term because of the huge developer community and extreme flexibility.
But it's currently really pain in the ass to find right settings, which deliver good results.
I'm going to print that! Your answer led me to search for the correct question, and I found an additional FAQ. Thank You Syntax for Stable Diffusion https://www.generativelabs.co/post/prompt-syntax-for-stable-diffusion-faqI see that the parts I was looking at are not f stops, but numerical weights. Also there's a 75 token limit. So some of the terribly long prompts I've seen could be truncated? Interesting. I'm using DALL-E and SDXL 1.0
Pages: 1 2 3 4 5 6 [7] 8 9 10 11 12 ... 180
|
Sponsors
Microstock Poll Results
Sponsors
|
Show Posts
